Internet in KGP: A Recent Take

- Kumar Abhishek, 5th Year Student, E & ECE Department

Internet is one of the most sought after lifelines in the KGP campus of today, much like the outside world. It has become a necessity in life where connections and contacts are the key to success, following food, clothing and shelter. Had a bad day at classes? Is the weather too humid? You don’t feel like eating mess food? But the internet is working? Alright then, Peace.

After returning to usual life back on campus this semester, I experienced on multiple occasions that internet at KGP isn’t as fast as it once used to be. Over the last five years, I remember switching between proxies 245, 247, 217, 211 and 212 to find the “best” speed proxy until a time came when it was replaced by a single unified proxy and re-structured WiFi networks. Feelings can be subjective, but numbers cannot. They always speak the truth. While I don’t have historical numbers / benchmarks to fully graph the trend of average internet speed, I am still sufficiently equipped to go ahead and do a benchmark of common scenarios that one encounters and leave you to make a comparison and form an opinion for yourself. All of the following tests have been performed on a system running Ubuntu 16.04 connected over the LAN.

Brief history on KGP’s internet links

Earlier KGP’s network used leased lines. It was then replaced with a single 2Gbps link internet from the National Knowledge Network (NKN), and it seems to have recently been upgraded. According to the section of “Bandwidth Utilization” on the CIC website , there are two 10Gbps internet links active at the moment provided by NKN, and each of the links is 15% utilized ( as of 31st August 3:45 AM ) – which means there is a 3Gbps connection from here to the outside world apart from the two gates on campus. It is quite evident that the infrastructure has improved and today the internet is in a better position than it was a few years ago.

Speedtest.net

Speedtest.net is one of those websites that rates the internet connections on 3 parameters – ping time ( the time it takes to send and receive something back from a server), download speed and upload speed. Showcased here are a few of the instances when I found myself on this website to test my beloved LAN connection.

23rd August, 1 AM, Nehru Hall LAN

lan1

24th August, 9 PM, Nehru Hall LAN

lan2

 

 

 

 

31st August, 1:30 AM, Nehru Hall LAN

lan3

 

 

 

 

The figures show that the speeds have infact improved over the course of a week, but a real-world test like downloading a large file shows the other side of the coin.

Large File Download – with and without download acceleration

Setting up new development environments, for instance, an SDK for developing Android apps, or a new framework, or a new version of Linux, or cloning a large Git repository usually requires one to download large files, usually of the order of gigabytes. So, this time around, I downloaded the Ubuntu 16.04 ISO image directly using the Chrome browser.

31st August, 1:30 AM

lfd

 

 

 

 

 

 

 

You can observe that I was consistently getting speeds below 100 KB/s, which means the connection speed was at a maximum of 0.8Mbps and it would have clearly taken the entire night for the download to finish. Using a download accelerator, Axel in my case, an equivalent to IDM which is quite popular among Windows users, this is what I observed:

axel

 

 

 

 

 

This is something unusual – in my over 4 year stay at KGP I have got speeds of about 150-200 KB/s without download accelerators consistently till date. On top of that, with a download accelerator, speeds used to go up to several MB/s – which meant that large files like these would be downloaded within 15 minutes almost certainly.

The curious case of YouTube

This semester, YouTube is perhaps the only website which hasn’t experienced a slowdown at all – rather it seem to have sped up. It is surprising that one can watch videos at 1080p, which buffer extremely fast (Flash would be proud) and play regularly without hiccups given the present scenario to low generic speeds. On right-clicking a YouTube video – there’s an aptly named option called “Stats for nerds”. Checking the option led to some figures showcased below. Featuring beside it is a system utility to measure the internet speed to give a better perspective.

youtube

 

 

 

 

 

 

 

 

According to YouTube, the link speed is 27Mbps – the streamed video is indeed 1080p. On observing the graph of the internet speed one can see the bursts as YouTube buffers the video over time. The very same internet speed that is around 1Mbps on average for large file downloads is able to stream videos at speeds of up to 4.0MB/s (= 32Mbps) for short bursts. A very sharp and strange contrast.

Conclusion

I don’t have any hard conclusions to draw at this point apart from the fact that the general speed levels have fallen sub par and that YouTube is an anomaly – it goes in the completely opposite direction in comparison to other websites. For casual internet browsing, speeds above a certain threshold are not so perceptible but while downloading large files, they definitely are. I leave the lot of you to form opinions based on what I’ve tried to convey apart from encouraging you to test link speeds and share your own experiences, especially with larger downloads.

I feel that one of the biggest issues with the KGP internet is the HTTP proxy itself – it leads to impediments in using various useful applications like Skype, Google Hangouts, and messaging apps like WhatsApp on the mobile without using a proxy app or a VPN service. For developers, blocking of SSH ports is a pain point. I agree that there are workarounds but wouldn’t it be better if they are not required at all? I also understand that the presence of a filter is necessary to prevent abuse of the network but there’s a balance to be found for ultimate cyber peace and harmony.

Leave a Comment

You must be logged in to post a comment.